Geophysical Journal International

Geophys. J. Int. (2013) doi: 10.1093/gji/ggt163

G
J
I
S
e
i
s
m
o
l
o
g
y
High-resolution relocation and mechanism of aftershocks of the 2007
Tocopilla (Chile) earthquake
A. Fuenzalida,
1
B. Schurr,
2
M. Lancieri,
3
M. Sobiesiak
4
and R. Madariaga
1
1
Laboratoire de G eologie CNRS-Ecole Normale Sup erieure, 75231 Paris Cedex 05, France. E-mail: fuenzalida@geologie.ens.fr
2
GFZ German Research Centre for the Geosciences, D-14473 Potsdam, Germany
3
Institut de Radioprotection et S uret e Nucl eaire, 92262 Fontenay-aux-Roses, France
4
Geophysiks, Universit at Kiel, Otto-Hahn-Platz 1, 24118 Kiel, Germany
Accepted 2013 April 18. Received 2013 April 18; in original form 2012 August 6
SUMMARY
We study the distribution of the aftershocks of Tocopilla M
w
7.7 earthquake of 2007 November
14 in northern Chile in detail. This earthquake broke the lower part of the seismogenic zone
at the southern end of the Northern Chile gap, a region that had its last megathrust earthquake
in 1877. The aftershocks of Tocopilla occurred in several steps: the rst day they were
located along the coast inside the co-seismic rupture zone. After the second day they extended
ocean-wards near the Mejillones peninsula. Finally in December they concentrated in the
South near the future rupture zone of the Michilla intermediate depth earthquake of 2007
December 16. The aftershock sequence was recorded by the permanent IPOC (Integrated
Plate Boundary Observatory in Chile) network and the temporary task force network installed
2 weeks after the main event. A total of 1238 events were identied and the seismic arrival
times were directly read from seismograms. Initially we located these events using a single
event procedure and then we relocated them using the double-difference method and a cross-
correlation technique to measure time differences for clusters of aftershocks. We tested a 1-D
velocity model and a 2-D one that takes into account the presence of the subducted Nazca
Plate. Relocation signicantly reduced the width of the aftershock distribution: in the inland
area, the plate interface imaged by the aftershocks is thinner than 2 km. The two velocity
models give similar results for earthquakes under the coast and a larger difference for events
closer to the trench. The surface imaged by the aftershocks had a length of 160 km. It extends
from 30 to 50 km depth in the northern part of the rupture zone; and between 5 and 55 km
depth near the Mejillones peninsula. We observed a change in the dip angle of the subduction
interface from 18

to 24

at a depth of 30 km. We propose that this change in dip is closely

associated with the upper limit of the rupture zone of the main event. We also studied the focal
mechanisms of the aftershocks, most of them were thrust events like the mainshock. As the
aftershock activity was signicantly reduced, on 2007 December 13, an M
L
6.1 event occurred
offshore of the Mejillones peninsula reactivating the seismicity. Three days later the Michilla
intraslab earthquake of M
w
6.8 ruptured an almost vertical fault with slab-push mechanism.
The aftershocks locations of this event dene a planar zone about 11 km in depth, situated
right bellow the subduction interface.
Key words: Earthquake source observations; Subduction zone processes; South America.
1 I NTRODUCTI ON
Northern Chile seismicity is dominated by the subduction of the
Nazca Plate beneath South America. Subduction in Chile generates
a large number of destructive earthquakes and tsunamis, like the
recent Maule (M
w
8.8) earthquake of 2010 February 27 (Vigny
et al. 2011) and the Valdivia megathrust earthquake of 1960 May
23 of magnitude M
w
9.5 (Kanamori &Cipar 1974; Cifuentes 1989).
In an attempt to understand the occurrence of large earthquakes in
the Chilean subduction zone several seismic gaps were identied
in the 1970s and 1980s by Lomnitz (1971), Kelleher et al. (1973)
and Compte et al. (1986). Gaps were dened as region that has not
had a large earthquake in a long time, generally more than 50 yr
for Chile. The geographical distribution and the extension of those
gaps were based on historical descriptions mainly based on the work
by Montessus de Balore (19111916) and are therefore affected by
large uncertainties. The availability of new seismic and geodetic
data has recently helped to identify more precisely the seismic
gaps and their boundaries (Ruegg et al. 2009; Madariaga et al.
2010).
C
The 2013. Authors Published by Oxford University Press on behalf of The Royal Astronomical Society 1
Geophysical Journal International Advance Access published May 17, 2013

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

2 A. Fuenzalida et al.
Figure 1. The Northern Chile seismic gap and the Tocopilla earthquake
of 2007. The Northern Chile gap is the region that was broken by the
M9 Iquique earthquake of 1877. We show the rupture zones of the largest
earthquakes of last century in the same area. The Tocopilla earthquake
occurred in the Southern part of the gap. The permanent seismic stations of
the IPOC network are shown with the green inverted triangles.
As shown in Fig. 1, one of the most conspicuous gaps in Chile is
the Northern Chile Gap, that extends for more than 500 kmfromthe
city of Arica in the north to the Mejillones peninsula in the south,
see for example Nishenko (1985). The last megathrust earthquake
in this gap dates back to 1877, an event whose magnitude has been
variously estimated between M
w
8.5 and 9 (Lomnitz 1971; Kausel
1986; Comte et al. 1986; Dorbath et al. 1990; Comte & Pardo
1991). The high rate of convergence of the Nazca Plate beneath the
South American Plate (6.7 cm yr
1
according to Angermann et al.
1999) and the lack of large earthquakes in the last 135 yr make this
area a very likely site for a future large earthquake. Currently the
gap is surveyed by the permanent Integrated Plate Boundary Obser-
vatory (IPOC) network of Northern Chile (www.ipoc-network.org)
deployed by Chilean, German and French researchers.
In the last 20 yr, two important earthquakes occurred just outside
the boundary of the Northern Chile seismic gap: the 2001 June 23,
M
w
8.4 Arequipa earthquake to the north and the 1995 July 30,
M
w
8.1 Antofagasta earthquake in the south. The 2001 Arequipa
earthquake broke the northern boundary of the gap and the con-
tiguous area, covering part of the rupture zone of the 1868 event
(Giovanni et al. 2002; Chlieh et al. 2011). The Antofagasta 1995
earthquake ruptured the region between 24.5

S and the Mejillones

peninsula at 22.5

S (Ruegg et al. 1996; Delouis et al. 1997). The

aftershock distribution (Delouis et al. 1997; Husen et al. 1999;
Nippress & Rietbrock 2007) as well as the geodetic observations
of this earthquake (Klotz et al. 1999; Pritchard et al. 2002; Chlieh
et al. 2004) show that the Mejillones peninsula stopped the rupture
transferring stresses into the southern part of the 1877 gap. In 2007,
the Tocopilla earthquake of magnitude M
w
7.7 broke approximately
20 per cent of the Northern Chile gap releasing only a small amount
of the accumulated stresses (Delouis et al. 2009; B ejar-Pizarro et al.
2010; Motagh et al. 2010; Peyrat et al. 2010). As shown in Fig. 1,
in the area situated immediately in the North of Tocopilla an M
w
7.4 event occurred on 1967 December 21, studied by Malgrange &
Madariaga (1983).
In the present paper, we study the aftershocks that followed the
2007 Tocopilla earthquake using 34 d of continuous recordings ac-
quired by the IPOC network and by the German task force (TF)
for earthquakes deployed 15 d after the mainshock (Sobiesiak et al.
2008) in the vicinity of the Mejillones peninsula. Because of the
excellent quality of the available data we were able to perform a
comprehensive analysis including event identication, accurate lo-
cation, magnitude estimation and the determination of focal mech-
anisms. Our aim is to determine the aftershock area activated by
the Tocopilla earthquake in order to study the following questions:
what were the up and down dip limits of the rupture? Can these
seismic observations dene the subducted plate geometry? Did the
Tocopilla earthquake activate upper-plate structures? And which are
the focal mechanisms that played the main role in the broken area?
2 THE TOCOPI LLA EARTHQUAKE
On 2007 November 14 the Tocopilla earthquake of magnitude M
w
7.7 (Delouis et al. 2009; Peyrat et al. 2010) broke the southern
part of the Northern Chile gap. The slip distribution of this earth-
quake, shown in Fig. 2, was obtained by kinematic inversion of the
source (Peyrat et al. 2010) using far and near eld data. The slip
distribution was composed of two main patches: the rupture started
in the northern patch and propagated towards the south breaking
a second patch that covers an area close to the ocean and stopped
at the Mejillones peninsula. The results of seismic inversion were
conrmed by interferometric synthetic aperture radar (InSAR) and
global positioning system (GPS) observations (B ejar-Pizarro et al.
2010; Motagh et al. 2010; Schurr et al. 2012). The joint inversion of
these geodetic data shows that the Tocopilla earthquake broke the
subduction interface between 30 and 50 km depth. This observation
is in agreement with the aftershock distribution of the rst 24 hr
following the Tocopilla earthquake (Delouis et al. 2009; Motagh
et al. 2010; Schurr et al. 2012).
Two important aftershocks of M
w
6.8 and M
w
6.3 occurred on
15 November 2007, almost 24 hr after the mainshock, off-shore the
Mejillones peninsula (Peyrat et al. 2010; Schurr et al. 2012). Both
are thrust events with the same mechanism as the Tocopilla earth-
quake. In the same zone we identied ve events of magnitude M
w
larger than 6 that occurred during the rst week after the Tocopilla
earthquake (Lancieri et al. 2012; Schurr et al. 2012) (see Fig. 2).
Seismic activity after November 15 occurred mainly offshore to
the north of the Mejillones peninsula until a strong intraslab event of
M
w
6.8 occurred at the bottom of the seismogenic interface on 2007
December 16. This event, the Michilla earthquake, was situated in-
side the Nazca Plate at 46 km of depth, 5 km below the subduction
interface. The Michilla earthquake and its main aftershocks had a
slab-push mechanism characterized by an along the slab compres-
sion. This event had an almost vertical main rupture plane (Peyrat
et al. 2010; Ruiz & Madariaga 2011). As discussed by Lemoine
et al. (2002) and Gardi et al. (2006) this kind of earthquake has also
been observed after other strong thrust events in Chile, Peru and
Mexico and their origin is still a matter of debate.

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 3
Figure 2. The M
w
7.7 Tocopilla earthquake of 2007 November 14. The
epicentre of the main event is shown with a red star and the black ellipses
show the slip patches of the main event determined by Peyrat et al. (2010).
The focal mechanisms from the global centroid moment tensor catalogue
show the main aftershocks of the event. Two shallow thrust aftershocks
occurred on 2007 November 15 offshore Mejillones peninsula. The slab-
push event of M
w
6.9 of 2007 December 16 is shown with a yellow star. The
stations of the task force (TF) seismic network that was deployed 2 weeks
after the mainshock are shown with magenta triangles. The green triangles
depict the location of the permanent IPOC network.
3 SEI SMOLOGI CAL DATA
The Tocopilla aftershocks were locally recorded by the IPOC and
TF networks in the North of Chile (Fig. 2). The IPOC network was
installed in 2006 thanks to an agreement between the International
Laboratory Montessus de Ballore (LIA) and the German Research
Centre for Geosciences (GFZ) with the aim of monitoring the seis-
mic gap of northern Chile. The IPOC network was composed of
12 stations equipped with broadband seismometers (STS-2) and
accelerometers (GMG-5 and Episensor FBA ES-T), spanning the
northern Chile gap with an inter-station distance of 80 km. The net-
work was operational in 2007 November when the main event and
the entire aftershock sequence were recorded at the seven southern-
most stations (Fig. 2). The PB04 station was located right above the
hypocentre of the main event and station PB05 was situated above
the Michilla slab-push event of 2007 December 16. The IPOC net-
work has been continuously improved and is currently operational
in Northern Chile. In order to understand the activity of the southern
part of the rupture, from 2007 November 29 a temporary network
TF was deployed around the Mejillones peninsula (Sobiesiak et al.
2008). The network was composed of 20 mostly short period in-
struments (L4-3D) recording continuously with a sampling rate of
100 Hz; the average distance between these stations was 15 km.
4 AFTERSHOCK LOCATI ONS
4.1 Event detection
Intense aftershock activity requires a sophisticated method for the
identication of events because their records often overlap in the
seismograms and the coda of the events raises the level of back-
ground noise. We chose to use the Filter Picker detection algorithm
of Lomax et al. (2012), which is an extension of the Allen (1978)
and Baer & Kradolfer (1987) methods. It is specically designed
to operate on continuous, real-time (but in this case we had off-line
data), broadband signals so as to avoid excessive triggering during
large events. We set the threshold parameters of Filter Picker so that
it could detect as many of the events of M
w
larger than 2 as possible.
The association procedure (binding) analyses the P wave arrival
times at every station to determine whether they are compatible with
propagation across the network from a common source (Lancieri
et al. 2011). This technique is based on the coincidence of a mini-
mum of ve picks within a given time window. We then automati-
cally cut windows of 3 min (60 s before and 120 s after the P phase)
around each of the events that were identied. During the rst 2
weeks, when only the IPOC network was available, we were not
able to locate all the detected events; this is due to the high seismic
activity and the large distance between the stations, the recorded P
phase were often hidden by the S phase or the coda of the previous
events. After the TF installation the distance between stations was
reduced, improving the detection capacity. Thanks to the stations
located on the Mejillones peninsula the offshore events were also
better located.
In order to obtain the best possible locations, we manually deter-
mined the arrival time of P and S phases on the seismic traces using
the seismic analysis code (SAC) of Goldstein et al. (2003). We also
read the P-phase polarities to determine focal mechanisms. Overall
we located 1238 events, 500 from November 14 to 28 when only
the IPOC network was available, and 738 from November 29 to
December 17 after the deployment of the TF network (see Table 1).
4.2 NonLinLoc locations
The precision of the hypocentre location depends on the quality and
number of phase readings, the spatial distribution of the stations and
a good knowledge of the velocity model. The good waveform data
of Northern Chile along with the manual reading of arrival times
permitted us to build an accurate catalogue of events. The events
were rst located with the NonLinLoc software (Lomax et al. 2000)
using two different velocity models (Fig. S1). We tested (1) a 1-D
velocity model derived by Husen et al. (2000) from the study of
the aftershock sequence of the 1995 Antofagasta earthquake, where
onshore and offshore instruments were available. And (2) a 2-D
velocity model inverted by Patzwahl et al. (1999) from the records
obtained during the CINCA seismic refraction project located just
north to the Mejillones peninsula.
The 1238 events located with the 1-D and 2-D velocity model
are shown in Figs 3(a) and (b), respectively. The epicentres de-
rived with either of the two velocity models are quite similar. The
aftershocks were mostly located on the two slip patches of the
main event plus a signicant amount of aftershocks situated off-
shore of the Mejillones peninsula (Figs 3a and b). The depths
of the events located using these two models were quite differ-
ent for the events located closer to the trench, as can be observed
in the vertical cross-sections plotted in Figs 3(c)(f). The after-
shocks of the Michilla earthquake of 2007 December 16 were lo-
cated very close to an almost vertical plane inside the Nazca Plate;
results are the same when using either one of the two structural
models.
As shown in Table 1, we divided the catalogue into three periods
of time according to differences in coverage and seismic activity:

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

4 A. Fuenzalida et al.
Table 1. Summary of depth errors for the two velocity models used to locate the aftershocks.
Period Network Date Located events Magnitude M
w
range Error 1-D (km) Error 2-D (km)
I IPOC 2007 November 142007 November 28 500 M
w
[2.67.8]
II IPOC + TF 2007 November 29-2007 December 15 487 M
L
[2.26.2] 1.73 2.15
III IPOC + TF 2007 December 162007 December 17 251 M
L
[1.86.8] 0.78 1,20
Figure 3. Hypocentral locations of the aftershocks of the Tocopilla earthquake of 2007 November 14 obtained with the Nonlinloc software. On the left-hand
side, we plot the locations obtained with the 1-D velocity model proposed by Husen et al. (2000). On the right-hand side, the locations determined using the
2-D model proposed by Patzwahl et al. (1999). Above the plan views we show EW cross-sections of the aftershocks. On (c) and (d), we plot the hypocentres
determined in period I when only the IPOC network was available. (d) and (f) are the locations obtained when both the Task Force and IPOC network were
operational.

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 5
Period I covers the rst 2 weeks of aftershocks of the Tocopilla
earthquake (cross-section on Figs 3c and d), when only seven IPOC
network stations were available; period II extends fromthe TFinstal-
lation until the occurrence of the Michilla earthquake on December
16 (20 additional stations); period III includes the Michilla earth-
quake and 2 d of its aftershocks. The events that occurred during
periods II and III were located using both the TF and IPOCnetworks
and are shown by the cross-sections of Figs 3(e) and (f).
The solutions obtained for the rst period using each of the two
velocity models provide a clear image of the subduction interface.
The events located offshore and in the southern area show a certain
dispersion due to poor azimuthal coverage by the IPOC network.
The 2-D model produced locations that were less scattered com-
pared to those obtained with the 1-D model but both have large
errors in depth. For periods II and III, after the installation of the
temporary TF network, the dispersion of hypocentre distribution
was signicantly reduced (Figs 3e and f). The hypocentres obtained
from the 1-D model (Fig. 3e) are distributed along a thin surface
with a dip angle that decreases as the events get closer to the trench.
The distribution of aftershocks located using the 2-D model images
a subduction interface that has a constant slope, so that the offshore
events reach the surface about 10 km from the trench.
To determine which velocity model provides the best locations,
we looked at the uncertainties of hypocentral depth. We focused on
depth because this is the parameter affected by the largest errors.
Table 1 summarizes the depth errors computed from the covariance
matrix derived for the two velocity models (Lomax et al. 2000).
As expected, the errors increase towards the trench for the two
models (Fig. S2). During period I, the network geometry dominated
the uncertainties, so that we were not able to decide which model
produced better locations based on the formal errors. For period II
(see Table 1), the errors in depth determination were improved with
a mean value of 1.73 km for the 1-D model and 2.15 km for the 2-D
model. Finally for period III, the errors were smaller because the
Michilla earthquake and its aftershocks were located inland, just
below the TF network. We obtained a mean error of 0.78 km for
the 1-D and 1.20 km for the 2-D models. In conclusion, the 1-D
model produce locations with smaller errors than the 2-D model,
but the difference between them may not be very signicant. The
inuence of the velocity model particularly on the depths of the
off-shore events is signicant and has to be kept in mind for the
interpretations.
4.3 Double-difference relocation
In order to improve the location of clusters of repeating after-
shocks, we used the double-difference (DD) relocation algorithm
(Waldhauser & Ellsworth 2000). The DD method relates travel-
time differences of pairs of nearby events recorded by the same
instrument to their spatial separations and therefore cancels out the
correlated errors arising from unmodelled structure along the seg-
ment of the path that they share. The DD method has been shown to
be particularly effective if precise differential traveltimes are mea-
sured using waveform cross correlation. Cross correlation-based
traveltime differences are often an order of magnitude more precise
than those derived from manually picked earthquake catalogues.
We focus our relocation analysis on periods II and III of the data set
starting from2007 November 29; these include the dense TF station
network. Cross correlations for P-phases were computed from the
vertical components after applying a 3-pole causal 110 Hz Butter-
worth bandpass lter for windows from0.5 s before to 1.5 s after the
P-wave pick. For S phases horizontal components were windowed
from1.0 s before to 2.0 s after the S-wave pick. For S phases the same
frequency lter as for P phases was used and in addition traces were
integrated to displacement because we found that this enhanced the
S signal. For the S phase windows, we stacked the cross-correlation
function of the two horizontal components and obtained the lag
time and correlation coefcient from the maximum amplitude of
the stack. Lag times with a correlation coefcient >0.7 were saved
for the relocation, using the correlation coefcient as a weight for
the hypocentre inversion. We calculated the cross-correlation func-
tion for event pairs with a maximum epicentral separation of 15 km.
This procedure yielded 27 117 cross correlation P lag time mea-
surements and 29 198 cross correlation S lag time measurements.
The event separation threshold was incrementally reduced to 5 km
during the relocation procedure. As velocity model we used the 1-D
model of Husen et al. (1999).
Fig. 4 shows the location obtained after the DD relocation. The
northern aftershocks were located on an elongated zone that extends
between 30 and 50 km depth (Fig. 4a). The aftershock zone is much
larger in the southern part as shown in Figs 4(b) and (c), where the
events were distributed between 5 and 55 km depth. In this area,
a change in the dip angle of the aftershock distribution is clearly
observed in the cross-section of Fig. 4(b). This change in dip angle
from 18

to 24

occurs at a depth of 30 km under the coastline. The

DD relocations reduce the width of the aftershock distribution to a
thin layer about 2 km of thickness that we identify as the subduction
interface. A comparison between locations before and after DD
relocation is presented in Fig. S3 of the supplementary material.
In Fig. 5 we show a zoom on the aftershock locations pointing out
some interesting features.
5 CHARACTERI ZATI ON OF
SEI SMI CI TY
5.1 Magnitude estimation
Once we built a catalog of aftershocks, we computed local magni-
tudes and focal mechanisms. For the rst period of data, Period I in
Table 1, a moment magnitude catalogue (M
w
) was computed using
the Brune (1970) spectral analysis of ground acceleration wave-
forms (Lancieri et al. 2011). That technique could not be applied to
the second period because the TF network had only short period in-
struments. Many events of this second period were recorded at only
two IPOC stations. For period II and III, when the largest number of
stations was available, we evaluated the local magnitude using the
Kanamori &Jennings (1978) procedure. This local magnitude (M
L
)
is dened as the logarithm of the maximum displacement measured
by a WoodAnderson torsion seismograph at an epicentral distance
of 100 km (Richter 1935):
M
L
= log A +log A
0
(1)
where A is one half the peak-to-peak displacement A
wa
weighted
by distance R by the relation: A = A
wa
R/100. A
0
is the reference
amplitude log A
0
(R = 100 km) = 3.
For all the short-period seismograms the mean value and linear
trend was removed before processing the data. The signals were
deconvolved from instrumental response and re-convolved with the
response of a WoodAnderson instrument. We determined the max-
imum amplitude of the S wave on the horizontal components and
corrected for the epicentral distance to evaluate the local magnitude

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

6 A. Fuenzalida et al.
Figure 4. Joint double-difference relocations applied to event pairs with a maximum epicentral separation of 15 km and waveform cross-correlation coef-
cients >0.7. On the left-hand side, we show the NS vertical cross-section of the relocated events for all the sequence. In the centre, we plot their geographical
distribution, plotting with red the relocated events that occurred during the rst 2 weeks and in blue those that took place after the TF installation. The EW
cross-sections (at the right-hand side) are plotted only for the second period because of their much better resolution. The aftershocks extend between 30 and
50 km depth in the northern area (a) and between 5 and 55 km (b and c) depth in the Mejillones peninsula zone. A dip change from 18

to 24

is observed
under the coastline at a depth of 30 km.
at each station. Finally, for each event we took the average of all
stations following the procedure proposed by Bobbio et al. (2009).
Fig. S4 of the supplementary material shows the number of events
as a function of local magnitude interval for the three periods dened
in Table 1. The rst period had the broader range of magnitudes with
several strong events : six events of magnitude larger than 6, and 33
larger than 5. Unfortunately, the geometry of the network did not
permit us to locate smaller events, especially in the southern and
offshore areas where the activity was the strongest. This situation
improved with the installation of the TF network, but this period
was less active with only one event with magnitude M
L
larger than
6, and six events of M
L
larger than 5. The third period corresponds
mainly to the aftershocks of the Michilla earthquake. In this period,
we had a very good azimuthal coverage necessary to locate a large
number of events, including those events with smaller magnitudes.
The b-value of the GutenbergRichter law was computed for the
periods I, II and III. Its value was close to 1 for the ranges [3.66],
[35] and [35], respectively (see Fig. S4).
With the goal of standardizing the two catalogues of magnitude
obtained for the two different geometries, we had to study the lower
magnitude thresholds. For magnitudes larger than 3 the effect of
changing the network on aftershock location was very strong. On
the other hand, if we choose a cut off magnitude of 4, the number
of events that appear in both catalogues becomes very small. For
this reason we made the spacetime analysis in Section 6 using only
local magnitudes M
L
greater than 3.5.
5.2 Focal mechanisms
We determined the focal mechanism of the aftershocks from the
rst motions of P waves read directly from the seismograms that
were used for the aftershock location in the previous sections. The
FPFIT software (Reasenberg et al. 1985) was used to compute
the focal mechanisms of the events having more than six polarity
readings. We read the polarities from the vertical seismograms at
each station and weighted it based on the signal to noise ratio. Using
the incidence angles obtained fromour locations, we determined the
focal mechanisms for epicentral distances less than 110 km. This
distance corresponds to the critical distance of the Pn phase arrival.
For period I, the number of readings and the distance between
stations was not enough to dene robust fault plane solutions. For
this reason we determined only the mechanisms of the events that
occurred after the TF network installation. Mechanisms for period
I were determined by Schurr et al. (2012) using moment tensors
determined from wave form modelling at regional distances.
We obtained 231 mechanisms in period II, before the Michilla
earthquake, with well resolved solutions, meaning errors of less
than 30

in strike, dip and rake (Fig. 6). Most of them are thrust
events just as the Tocopilla mainshock, and a signicant part of
them were located offshore. Unfortunately, in the offshore area the
azimuthal coverage is less good so that the mechanisms are less
well resolved. Inland, in the deeper seismogenic zone we observe a
larger variability in the focal mechanism, including a fewalong-slab

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 7
Figure 5. Detailed image of the cross-section under the Mejillones peninsula. In (a) relocated events show that some events split from the interface delineating
a possible splay fault; in (b) at the depth of the kink we observe a small branch that penetrates into the Nazca Plate; and in (c) the aftershocks of Michilla
earthquake broke the oceanic crust of the Nazca Plate along an almost vertical plane that penetrates 11 km into the subducted slab. In the background of the
main cross-section in the top frame we plot the receiver function section computed by Sodoudi et al. (2011), migrated using the 1-D velocity model of Husen
et al. (2000). The strong red phase is interpreted as the oceanic Moho, so that the Michilla earthquake seems to have broken the entire oceanic crust.
tensional mechanisms (slab-pull) such as the M
L
4.6 event of 2007
December 15 (Fig. 6). A short sequence of aftershocks started with
the M
L
6.1 thrust event of 2007 December 13; this event triggered
several M
L
5 events for which we could determine their mechanisms.
The rose diagram in Fig. 7 summarizes the fault angles computed
for events with dip errors less than 20

. We observe that most of the

events had dips between 15

and 30

. We see no evidence of non

thrust events in our catalogue as those found by Schurr et al. (2012)
at shallow depths in the upper plate; those events occurred during
he rst week of aftershock when the TF was not yet in operation.
The focal mechanismof the Michilla earthquake (in red in Fig. 7)
had an along the slab compression mechanism (slab-push). As
shown in the gure this event occurred on an almost vertical plane
inside the Nazca Plate at 46 km depth. Its aftershocks (plotted in
red) and foreshocks (in blue) are shown in Fig. 8. The foreshocks
occurred during the 2 weeks that preceded the Michilla event and
had moderate magnitude (M
L
3) with a similar slab-push mech-
anism to Michilla event. In the rose diagram presented as an inset
in Fig. 7 we observe that almost all the aftershoks had a rake angle
close to 90

. We also found two preferred angles of dip: a steeper

one close to 80

similar to that of the slab-push Michilla earthquake,

and a second dip angle close to 25

that is typical of the thrust events

of the main Tocopilla event. Thus inside the aftershock zone of the
Michilla earthquake most events have the same mechanism as the
slab-push event, but there are still a few shallow dip slip thrust
events.They can be seen in Fig. S4.
6 SPACE TI ME DI STRI BUTI ON
The aftershock sequence of the Tocopilla earthquake is charac-
terized by three main zones of activity: the area broken by the
mainshock located inland; a strong activity off the Mejillones penin-
sula and the fore- and aftershock sequence of Michilla event. The
temporal evolution of the events including their magnitude are
shown separately for the two areas in Fig. 8: inland (plotted in
red) and offshore (blue).
For the events located inland the activity started with the main
Tocopilla event shown with a large star and its aftershocks were of
moderate magnitude. Two patches of seismicity are highlighted: the
events associated with the mainshock rupture area and the seismicity
that was triggered offshore of the Mejillones peninsula producing
two events of M
w
6.8 and another one of M
w
6.3 are indicated
with the left-hand arrow. Seismicity remained high during the rst
week, with several events of magnitude greater than 6. It was re-
activated several times and eventually became quieter during the
second and third week of aftershocks. Finally, the offshore area was
reactivated by the M
L
6.1 event that occurred on 2007 December
13. Some events migrated to the deeper parts of the seismogenic
zone during the 3 d that preceded the Michilla earthquake of 2007
December 16.
7 DI SCUSSI ON
The 2007, Tocopilla earthquake is one of the best instrumented
earthquakes in Chile. Thanks to the IPOC network installed before
the earthquake, we were able to study the principal features of the
mainshock and its larger aftershocks (Peyrat et al. 2010; Schurr
et al. 2012). Northern Chile offers ideal conditions for the study
of seismicity because site effects are very limited and attenuation
is low as discussed by Lancieri et al. (2012). In the 2 weeks that
followed the main event, most of the aftershocks occurred offshore
and in the southern part of the main rupture, where there were fewer

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 A. Fuenzalida et al.
Figure 6. Focal mechanism of the aftershocks of the period from 2007 November 29 and 2007 December 16 when the TF network was recording. The rose
diagram in the upper right-hand side shows the dip angles of the fault planes which are clearly dominated by an angle of 18

. The events of magnitude larger

than 4.5 are highlighted in black and events of smaller magnitude are shown in grey. The Michilla (slab-push) and the M
L
6.1 event of 2007 December 13
epicentres are represented by stars.
stations installed before the earthquake. For this reason, we were
not able to locate all the events that were identied by the automatic
event picking algorithm. This situation improved signicantly once
the TF network was installed. The quantity of instruments and its
spatial distribution allowed us to better identify seismic arrivals and
reduce the error of locations particularly of southern and offshore
events.
The location of the aftershocks situated offshore of the
Mejillones peninsula was signicantly affected by the velocity
model used for the determination of hypocentres (see Fig. 9). We
tested two velocity models that had been proposed for the region
by earlier studies. A 1-D model determined for the aftershocks of
the 1995 Antofagasta earthquake by Husen et al. (1999), and a 2-D
model obtained by Patzwahl et al. (1999) for a seismic prole at
the latitude of Tocopilla. The main difference in the locations is
in the depth distribution of offshore events: the locations obtained
using the 1-D model dene a surface that smoothly joins the trench.
Locations obtained with the 2-D model, on the other hand, dene a
plane that intersects the seaoor 10 km inland from the trench. We
think that the 1-D velocity performs better for this zone because the
errors were smaller, nevertheless it is less realistic because it lacks
the slab. Locations obtained with the 1-D model dene a dip angle
of the plate interface that is in very good agreement with previous
studies of the seismicity in the vicinity of Antofagasta, that imaged
a plate interface zone with dip angles in the range from 17

to 18

(Comte et al. 1994; Delouis et al. 1996).

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 9
Figure 7. Focal mechanisms of the aftershocks of the Michilla earthquake of 2007 December 16 (in red) and seven of its foreshocks that occurred near its
fault plane (in blue). Grey circles represents the background seismicity. Triangles and diamonds represents the nearby seismic stations.The rose diagram to the
right-hand side of the gure shows that there were two dominant mechanisms during this period: thrust events with dip angles between 1

and 30

and slab
compression events (slab push) with dip angles close to 60

.
Relocation using cross correlation of clusters of aftershocks sig-
nicantly reduced the width of the seismogenic structures; after-
shocks collapsed into a very thin surface of a thickness of about
2 km that we interpret as the shear zone located in the immediate
vicinity of the plate interface. In Fig. 5, we zoom on the cross-
section of Fig. 4(b). Three main features are evident: a shallow
structure appears to detach from the subduction interface (Fig. 5 a)
in the offshore zone; in the region of the kink, shown in Fig. 5(b),
we observe that the hypocentres do not delineate a single layer but
a small branch appears near 70.5

. Finally, as shown in Fig. 5(c),

the Michilla earthquake rupture zone is clearly delimited by its af-
tershocks that dene an 11 km depth section situated immediately
below the subduction interface.
The kink found from the 1-D locations across the Mejillones
peninsula increases the dip angle by about 6 degrees as observed in
Fig. 5. Unlike the study of Contreras-Reyes et al. (2012), the change
in dip angle of our aftershock distribution is situated at 30 kmdepth,
instead of 18 km as proposed by them. Our locations do not image
an abrupt change of dip but rather a broad zone of about 10 km
width in the EW direction (see Fig. 5 b). We observe the presence
of a small seismicity branch inside the Nazca Plate right under the
kink zone. The location of this change in dip corresponds to the
upper limit of the aftershock distribution of the northern part of
the Tocopilla earthquake. It is tempting to associate this kink with
the arrest of the rupture, but other authors (Schurr et al. 2012) have
suggested that the main event was stopped by a change in frictional
properties of the plate interface. We have no data that is relevant
to resolve these two hypotheses, although the change in frictional
properties may be associated with the change in dip of the plate
interface. To improve the geometry of the plate interface near the
trench we would need ocean bottom instruments in order to resolve
variations on the dip angle using the same localization methods.
In the inland zone, we did not nd any evidence of activity at shal-
low depths, as could have been expected for some recent normal
faults in the Mejillones peninsula plotted in Fig. 10 after (Armijo &
Thiele 1990; Victor et al. 2011; Vargas et al. 2011). We observe a
strong seismic activity under the Mejillones peninsula below these
faults, but these events have depths of about 30 km, so that they are
close to the plate interface. The only evidence of shallow seismic-
ity is in the offshore area of the Mejillones peninsula: a cluster of
aftershocks located in the upper plate that may be due to a possible
splay fault (see Fig. 5 a). As we have already discussed those events
are not sufciently well located because of the lack of ocean bottom
instruments. Another interesting observation is the diagonal align-
ment of seismicity just north of Mejillones peninsula in the offshore
area. This alignment is not related to known geological structures
and could be interpreted as due the presence of uids as was pro-
posed by Nippress & Rietbrock (2007) for the aftershocks of the
Antofagasta earthquake, but this hypothesis needs to be conrmed
by other observations.
The focal mechanism of almost all well located events was thrust
faulting, typical of subduction events. A signicant part of these
events was located offshore where fault plane solutions are not
well constrained due to poor azimuthal coverage. For period I,

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

10 A. Fuenzalida et al.
Figure 8. Migration of seismicity during the Tocopilla aftershocks of 2007. The catalogue is separated into inland (red) and offshore (blue) zones. On the
right-hand side, we plot the spacetime evolution of seismicity, the ordinate represents the latitude and the abscissa the time in days. The gure shows the
patches of seismicity for the mainshock (the large yellow star in right-hand panel) in agreement with the slip distribution of Fig. 2. This main event triggered the
offshore events (black arrow on the left-hand side). The Mejillones offshore seismicity (plotted in blue) remained active longer than the main rupture plotted in
red. Activity decreased after the rst 2 weeks to restart 3 d before the Michilla earthquake with a 6.1 thrust event that preceded the Michilla event (black arrow
to the right-hand side). Events of M
L
> 6 are shown with yellow stars.
Figure 9. The aftershock distribution of the 2007 Tocopilla earthquake superimposed on the velocity model recently proposed by Contreras-Reyes et al. (2012)
for the latitude 22

S. Our locations are centred on the Mejillones peninsula. On the left-hand side, we plot the aftershock locations obtained with a 1-D layered
velocity model proposed by Husen et al. (1999). On the right-hand side, locations determined using the 2-D model proposed by Patzwahl et al. (1999). The
velocity model proposed by Contreras-Reyes et al. (2012) is well dened above the the dashed-line. The two aftershock distributions are very similar at depth,
delineating the same planar zone situated near the top of the oceanic crust of the subducted Nazca Plate. At shallower depths, the locations differ signicantly
but we consider that these aftershocks were not well located because of the lack of ocean bottom instruments.
Schurr et al. (2012) found some normal fault mechanisms in the
upper plate from waveform modelling, but we could not conrm
their results because fault plane solution were not well constrained
in the southern part of the aftershock sequence during the two rst
weeks. Although we looked carefully for shallow aftershocks in our
data we could not nd a single well located event in the region
close to the coast. This is in stark contrast with the seismicity pro-
duced by the Maule M
w
8.8 earthquake of 2010 February 27 whose

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 11
Figure 10. 1-D Locations in the Mejillones peninsula zone. This penin-
sula has well-known geological structures plotted on the map (Armijo &
Thiele 1990). We did not nd any shallow aftershocks associated with these
structures.
largest aftershock was an event that occurred on a shallow fault
near Pichilemu (Faras et al. 2011). Although the aftershocks or
the Maule earthquake have not been fully relocated yet, the studies
by Lange et al. (2012) and Rietbrock et al. (2012) have reported
many shallow and quite a few deep aftershocks. The rupture zone
delineated by the aftershock distribution seems to be much wider
in Maule than in Tocopilla. Part of the difference may simple be
that a very large event like Maule necessarily breaks a broad zone
of the plate interface and produces many aftershocks away from
this seismogenic zone. In Tocopilla, we observe that the aftershocks
occurred in a very thin zone at the contact between the Nazca and
South American plates. It is possible that the localization of after-
shocks in a narrow zone is very different for megathrust events and
regular subduction events like Tocopilla.
The Michilla slab-push earthquake and its aftershocks occurred
near the centre of the network so that their focal mechanisms are
very well determined. The rupture zone of this event was an almost
vertical, planar zone very well dened by the aftershocks. These
slab-compression events have been observed several times in Chile
(Astiz & Kanamori 1986; Lemoine et al. 2002; Gardi et al. 2006),
but this is the rst time that the rupture plane is so well delineated
by the distribution of the aftershocks. The majority of these after-
shocks had very similar mechanisms as those of the main event.
They ruptured a compact zone of 14 km length by 11 km depth with
the hypocentre situated in the centre. This rupture zone extends
from the top of the megathrust seismogenic zone to a depth that is
probably very close to the base of the oceanic crust. In Fig. 5, we
compare our locations with the results of a receiver functions sec-
tions obtained by Sodoudi et al. (2011). This slab push earthquake
occurred below the second patch of the mainshock and was very
likely triggered by stress transfer from the main shock as discussed
by Peyrat et al. (2010). It occurred 3 d after a reactivation of the af-
tershock seismicity offshore the Mejillones peninsula as observed
in Fig. 8. This reactivation occurred when the aftershocks of the
Tocopilla main event had signicantly diminished. The Michilla
earthquake had its own set of aftershocks and foreshocks that re-
mained very close to the rupture zone of the earthquake at least
during the 2 d that we studied. Of all slab-push events in Chile, the
Punitaqui earthquake of 1997 October 15 (Lemoine et al. 2002;
Gardi et al. 2006) is the most similar to Michilla, at least in size and
position with respect to the transition zone in Chile. As for Michilla,
the Punitaqui earthquake was preceded by a series of thrust events
near the centre of the seismogenic zone on the plate interface. Those
events, however, occurred as a swarm in 1997 July at a distance of
more than 50 km from the Punitaqui hypocentre. For both events we
claim that stress reduction on the plate interface loaded the region
near the tip of the transition zone, where slip occurs continuously,
triggering these slab push earthquakes inside the subducted Nazca
Plate.
8 CONCLUSI ONS
We studied the aftershock distribution of the 2007, M
w
7.7 To-
copilla earthquake of Northern Chile. We used data from a sparse
permanent network of stations (IPOC) and a dense temporary array
(TF) to locate the aftershocks, determine their magnitude and iden-
tify their mechanisms. The aftershocks dene an activated zone of
about 160 km of length in north-south direction by 25 km of width
in the northern area near the down-dip limit of the seismogenic
zone, and a broad zone in the southern part of the rupture of about
50 km of large. Using P and S phases individually read from seis-
mograms and a double-difference method, we could signicantly
improve the locations. We found that the aftershocks of the To-
copilla earthquake are distributed on a very narrow seismogenic
zone at the interface between the Nazca and South American plates.
The vertical cross-section of the seismicity depends signicantly on
the velocity model used for the locations. The simpler 1-D model
produces a curved interface that smoothly reaches the trench, while
a 2-D velocity model derived from a local seismic experiment pro-
duces a seismogenic interface that intersects the surface far from
the trench. We can not resolve the differences between these two ve-
locity models because the off-shore locations are less accurate than
the inland ones. To improve the quality of the offshore locations
and better image the geometry of the slab we would have required
ocean bottominstruments. Our study conrms the previous work by
Peyrat et al. (2010); B ejar-Pizarro et al. (2010) who proposed that
the Tocopilla earthquake broke the bottomof the plate interface, just
above the transition zone. The plate interface in the Tocopilla region
did not have a large earthquake in at least 135 yr since the great
Iquique earthquake of 1977. A possible scenario is that the rupture
of the bottom part of the seismogenic zone may have increased the
risk of a future tsunamigenic earthquake that will break the entire
plate interface. We cannot exclude however an alternative scenario
in which a series of earthquakes of magnitude close to 8 may occur
along the coast of the northern Chile gap delaying the occurrence of
a very large megathrust earthquake. Future observations will permit
us to decide which of these two scenarios is more likely in Northern
Chile.

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

12 A. Fuenzalida et al.
ACKNOWLEDGEMENTS
This research was carried out under the Montessus de Ballore In-
ternational Laboratory established between the University of Chile
and the Centre National de la Recherche Scientique (CNRS) in
France. This work was supported by FONDECYT No. 1100429
in Chile and by ANR project S4 in France. AF was supported
by a fellowship from AXA. We thank the many researchers in
Chile, Germany and France that have installed and maintained seis-
mic networks in Northern Chile, in particular GFZ that makes the
IPOC data openly available from the GEOFON website.We deeply
thank Sergio Ruiz and Claudio Satriano for numerous discussions.
The manuscript was improved thanks to the constructive comments
of the editor and two anonymous referees. We used the Generic
Mapping Tool (http://gmt.soest.hawaii.edu/) to prepare many of the
gures.
REFERENCES
Allen, R., 1978. Automatic earthquake recognition and timing from single
traces, Bull. seism. Soc. Am., 68(5), 15211532.
Angermann, D., Klotz, J. & Reigber, C., 1999. Space-geodetic estimation
of the Nazca-south America Euler vector, Earth planet. Sci. Lett., 171(3),
329334.
Armijo, R. & Thiele, R., 1990. Active faulting in northern Chile: ramp
stacking and lateral decoupling along a subduction plate boundary?, Earth
planet. Sci. Lett., 98(1), 4061.
Astiz, L. & Kanamori, H., 1986. Interplate coupling and temporal variation
of mechanisms of intermediate-depth earthquakes in Chile, Bull. seism.
Soc. Am., 76(6), 16141622.
Baer, M. & Kradolfer, U., 1987. An automatic phase picker for local and
teleseismic events, Bull. seism. Soc. Am, 77(4), 14371445.
B ejar-Pizarro, M. et al., 2010. Asperities and barriers on the seismogenic
zone in North Chile: state-of-the-art after the 2007 Mw 7.7 Tocopilla
earthquake inferred by GPS and InSAR data, Geophys. J. Int., 183(1),
390406.
Bobbio, A., Vassallo, M. & Festa, G., 2009. A local magnitude scale for
southern Italy, Bull. seism. Soc. Am., 99(4), 24612470.
Brune, J., 1970. Tectonic stress and the spectra of seismic shear waves from
earthquakes, J. geophys. Res., 75(26), 49975009.
Chlieh, M., De Chabalier, J., Ruegg, J., Armijo, R., Dmowska, R., Cam-
pos, J. & Feigl, K., 2004. Crustal deformation and fault slip during the
seismic cycle in the North Chile subduction zone, from GPS and InSAR
observations, Geophys. J. Int., 158(2), 695711.
Chlieh, M. et al., 2011. Interseismic coupling and seismic potential along
the Central Andes subduction zone, J. geophys. Res., 116(B12), B12405,
doi:10.1029/2010JB008166.
Cifuentes, I., 1989. The 1960 Chilean earthquakes, J. geophys. Res., 94(B1),
665680.
Comte, D. & Pardo, M., 1991. Reappraisal of great historical earthquakes
in the northern Chile and southern Peru seismic gaps, Nat. Hazards, 4(1),
2344.
Comte, D., Eisenberg, A., Lorca, E., Pardo, M., Ponce, L., Saragoni, R.,
Singh, S. & Su arez, G., 1986. The 1985 central Chile earthquake: a
repeat of previous great earthquakes in the region?, Science, 233(4762),
449453.
Comte, D., Pardo, M., Dorbath, L., Dorbath, C., Haessler, H., Rivera, L.,
Cisternas, A. & Ponce, L., 1994. Determination of seismogenic interplate
contact zone and crustal seismicity around Antofagasta, northern Chile
using local data, Geophys. J. Int., 116(3), 553561.
Contreras-Reyes, E., Jara, J., Grevemeyer, I., Ruiz, S. & Carrizo, D., 2012.
Abrupt change in the dip of the subducting plate beneath north Chile,
Nat. Geosci., 5(5), 342345.
Delouis, B., Cisternas, A., Dorbath, L., Rivera, L. & Kausel, E., 1996. The
Andean subduction zone between 22 and 25 S (northern Chile): precise
geometry and state of stress, Tectonophysics, 259(13), 81100.
Delouis, B. et al., 1997. The Mw = 8.0 Antofagasta (northern Chile) earth-
quake of 30 July 1995: a precursor to the end of the large 1877 gap, Bull.
seism. Soc. Am., 87(2), 427445.
Delouis, B., Pardo, M., Legrand, D. & Monfret, T., 2009. The Mw 7.7
Tocopilla earthquake of 14 November 2007 at the southern edge of the
northern Chile seismic gap: rupture in the deep part of the coupled plate
interface, Bull. seism. Soc. Am., 99(1), 8789.
Dorbath, L., Cisternas, A. & Dorbath, C., 1990. Assessment of the size
of large and great historical earthquakes in Peru, Bull. seism. Soc. Am.,
80(3), 551576.
Faras, M., Comte, D., Roecker, S., Carrizo, D. & Pardo, M.,
2011. Crustal extensional faulting triggered by the 2010 Chilean
earthquake: the Pichilemu seismic sequence, Tectonics, 30(6),
doi:10.1029/2011TC002888.
Gardi, A., Lemoine, A., Madariaga, R. & Campos, J., 2006. Modeling of
stress transfer in the Coquimbo region of central Chile, J. geophys. Res.,
111, B04307, doi:10.1029/2004JB003440.
Giovanni, M., Beck, S. & Wagner, L., 2002. The June 23, 2001 Peru earth-
quake and the southern Peru subduction zone, Geophys. Res. Lett., 29(21),
141.
Goldstein, P., Dodge, D., Firpo, M. & Minner, L., 2003. SAC2000: sig-
nal processing and analysis tools for seismologists and engineers, Int.
Geophys., 81, 16131614.
Husen, S., Kissling, E., Flueh, E. & Asch, G., 1999. Accurate hypocentre
determination in the seismogenic zone of the subducting Nazca Plate in
northern Chile using a combined on-/offshore network, Geophys. J. Int.,
138(3), 687701.
Husen, S., Kissling, E. & Flueh, E., 2000. Local earthquake tomography
of shallow subduction in north Chilea combined onshore and offshore
study, J. geophys. Res., 105(B12), 28 18328 198.
Kanamori, H. & Cipar, J., 1974. Focal process of the great Chilean earth-
quake May 22, 1960, Phys. Earth planet. Inter., 9(2), 128136.
Kanamori, H. & Jennings, P., 1978. Determination of local magnitude, ML,
fromstrong-motion accelerograms, Bull. seism. Soc. Am., 68(2), 471485.
Kausel, E., 1986. Los terremotos de agosto de 1868 y mayo de 1877 que
afectaron el sur del per u y norte de chile, Boletn de la Academia Chilena
de Ciencias, 3, 814.
Kelleher, J., Sykes, L. & Oliver, J., 1973. Possible criteria for predicting
earthquake locations and their application to major plate boundaries of
the pacic and the caribbean, J. geophys. Res., 78(14), 25472585.
Klotz, J. et al., 1999. GPS-derived deformation of the Central Andes includ-
ing the 1995 Antofagasta Mw = 8.0 earthquake, Pure appl. Geophys.,
154(34), 709730.
Lancieri, M., Fuenzalida, A., Ruiz, S. & Madariaga, R., 2011. Magnitude
scaling of early-warning parameters for the Mw 7.8 Tocopilla, Chile,
earthquake and its aftershocks, Bull. seism. Soc. Am., 101(2), 447463.
Lancieri, M., Madariaga, R. & Bonilla, F., 2012. Spectral scaling of the
aftershocks of the Tocopilla 2007 earthquake in northern Chile, Geophy.
J. Int., 189(1), 469480.
Lange, D. et al., 2012. Aftershock seismicity of the 27 February 2010 Mw
8.8 Maule earthquake rupture zone, Earth planet. Sci. Let., 317(318),
413425.
Lemoine, A., Madariaga, R. & Campos, J., 2002. Slab-pull and slab-push
earthquakes in the Mexican, Chilean and Peruvian subduction zones,
Phys. Earth planet. Int., 132(1), 157175.
Lomax, A., Virieux, J., Volant, P. & Berge, C., 2000. Probabilistic earth-
quake location in 3D and layered models: introduction of a Metropolis-
Gibbs method and comparison with linear locations, in Advances in Seis-
mic Event Location, pp. 101134, eds, Thurber, C.H. & Rabinowitz, N.,
Kluwer, Amsterdam.
Lomax, A., Satriano, C. & Vassallo, M., 2012. Automatic picker develop-
ments and optimization: lterpickera robust, broadband picker for real-
time seismic monitoring and earthquake early warning, Seism. Res. Lett.,
83(3), 531540.
Lomnitz, C., 1971. Grandes terremotos y tsunamis en Chile durante el
periodo 1535-1955, Geos. Panamericana, 1, 151171.
Madariaga, R., M etois, M., Vigny, C. & Campos, J., 2010. Central Chile
nally breaks, Science, 328(5975), 181182.

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Tocopilla 2007 aftershocks 13
Malgrange, M. & Madariaga, R., 1983. Complex distribution of large thrust
and normal fault earthquakes in the Chilean subduction zone, Geophys.
J. R. astr. Soc., 73(2), 489505.
Montessus de Balore, F., 19111916. Historia sismica de los Andes
meridionales al sur del paralelo XVI, Vol. 6, Cervantes, Santiago de
Chile.
Motagh, M., Schurr, B., Anderssohn, J., Cailleau, B., Walter, T., Wang, R.
& Villotte, J., 2010. Subduction earthquake deformation associated with
14 November 2007, Mw 7.8 Tocopilla earthquake in Chile: results from
InSAR and aftershocks, Tectonophysics, 490(1), 6068.
Nippress, S. & Rietbrock, A., 2007. Seismogenic zone high permeability in
the Central Andes inferred from relocations of micro-earthquakes, Earth
planet. Sci. Lett., 263(34), 235245.
Nishenko, S., 1985. Seismic potential for large and great interplate earth-
quakes along the Chilean and southern Peruvian margins of South
America: a quantitative reappraisal, J. geophys Res., 90(B5), 3589
3615.
Patzwahl, R., Mechie, J., Schulze, A. & Giese, P., 1999. Two-dimensional
velocity models of the Nazca plate subduction zone between 19.5 S and 25
S from wide-angle seismic measurements during the CINCA95 project,
J. geophys. Res., 104(B4), 72937317.
Peyrat, S., Madariaga, R., Buforn, E., Campos, J., Asch, G. & Vilotte, J.,
2010. Kinematic rupture process of the 2007 tocopilla earthquake and its
main aftershocks from teleseismic and strong-motion data, Geophys. J.
Int., 182(3), 14111430.
Pritchard, M., Simons, M., Rosen, P., Hensley, S. & Webb, F., 2002. Co-
seismic slip from the 1995 July 30 Mw = 8.1 Antofagasta, Chile, earth-
quake as constrained by InSAR and GPS observations, Geophys. J. Int.,
150(2), 362376.
Reasenberg, P. & Oppenheimer, D., 1985. FPFIT, FPPLOT, and FPPAGE:
fortran computer programs for calculating and displaying earthquake
fault-plane solutions, US Dept. of the Interior, Geological Survey.
Richter, C., 1935. An instrumental earthquake magnitude scale, Bull. seism.
Soc. Am., 25(1), 132.
Rietbrock, A., Ryder, I., Hayes, G., Haberland, C., Comte, D., Roecker,
S. & Lyon-Caen, H., 2012. Aftershock seismicity of the 2010
Maule Mw = 8.8, Chile, earthquake: correlation between co-seismic
slip models and aftershock distribution?, Geophys. Res. Lett., 39(8),
doi:10.1029/2012GL051308.
Ruegg, J. et al., 1996. The Mw =8.1 Antofagasta (North Chile) earthquake
of July 30, 1995: rst results fromteleseismic and geodetic data, Geophys.
Res. Lett, 23(9), 917920.
Ruegg, J. et al., 2009. Interseismic strain accumulation measured by GPS
in the seismic gap between Constituci on and Concepci on in Chile, Phys.
Earth planet. Int., 175(12), 7885.
Ruiz, S. &Madariaga, R., 2011. Determination of the friction lawparameters
of the Mw 6.7 Michilla earthquake in northern Chile by dynamic inver-
sion, Geophys. Res. Lett., 38(9), L09317, doi:10.1029/2011GL047147.
Schurr, B., Asch, G., Rosenau, M., Wang, R., Oncken, O., Barrientos, S.,
Salazar, P. & Vilotte, J., 2012. The 2007 M7. 7 Tocopilla northern Chile
earthquake sequence: implications for along-strike and downdip rup-
ture segmentation and megathrust frictional behavior, J. geophys. Res.,
117(B5), B05305, doi:10.1029/2011JB009030.
Sobiesiak, M. et al., 2008. The m 7.7 tocopilla earthquake and its aftershock
sequence: deployment of a task force local network, EOS Trans. AGU,
89(23), Jt. Assembly Suppl. Abstract S24A-04.
Sodoudi, F., Yuan, X., Asch, G. & Kind, R., 2011. High-resolution
image of the geometry and thickness of the subducting Nazca
lithosphere beneath northern Chile, J. geophys. Res., 116, B04302,
doi:10.1029/2010JB007829.
Vargas, G., Palacios, C., Reich, M., Luo, S., Shen, C.-C., Gonz alez, G. &
Wu, Y.-C., 2011. U-series dating of co-seismic gypsum and submarine
paleoseismology of active faults in northern Chile (23 s), Tectonophysics,
497(1), 3444.
Victor, P., Sobiesiak, M., Glodny, J., Nielsen, S. & Oncken, O., 2011. Long-
term persistence of subduction earthquake segment boundaries: evidence
from Mejillones Peninsula, northern Chile, J. geophys. Res., 116(B2),
B02402, doi:10.1029/2010JB007771.
Vigny, C. et al., 2011. The 2010 Mw 8.8 maule megathrust earthquake of
Central Chile, Monitored by GPS, Science, 332(6036), 14171421.
Waldhauser, F. & Ellsworth, W., 2000. A double-difference earthquake lo-
cation algorithm: method and application to the northern Hayward fault,
California, Bull. seism. Soc. Am., 90(6), 13531368.
SUPPORTI NG I NFORMATI ON
Additional Supporting Information may be found in the online ver-
sion of this article:
Figure S1. P-wave velocity model used for the location of the 2007
Tocopilla aftershocks.
Figure S2. Error distibution of aftershock locations of the 2007
Tocopilla earthquake.
Figure S3. Aftershocks of the 2007 Tocopilla earthquake.
Figure S4. Determination of the GutenbergRichter law for the
Tocopilla aftershocks.
Figure S5. Zoom on the aftershock zone of the 2007 December
16 Michilla earthquake (http://gji.oxfordjournals.org/lookup/suppl
/doi:10.1093/gji/ggt163/-/DC1).
Please note: Oxford University Press is not responsible for the con-
tent or functionality of any supporting materials supplied by the
authors. Any queries (other than missing material) should be di-
rected to the corresponding author for the article.

a
t

E
c
o
l
e

n
o
r
m
a
l
e

s
u
p

r
i
e
u
r

o
n

M
a
y

1
9
,

2
0
1
3
h
t
t
p
:
/
/
g
j
i
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m